Temperature and solvent-polarity dependence of the absorption and

Michael J. Frampton, Huriye Akdas, Andrew R. Cowley, Joy E. Rogers, Jonathan E. Slagle, Paul A. Fleitz, Mikhail Drobizhev, Aleksander Rebane, and Harr...
0 downloads 0 Views 1MB Size
J . Phys. Chem. 1989, 93, 5437-5444 radical anion.l1.l2 The similarity of the v7? frequency of the trianion radical to that of the semiquinone anion indicates that these bonds are of similar strength. While protonation of the carboxylate oxygen has little effect on the ring or C-X (X = COz or COZH) bonds, protonation of p-benzosemiquinone anion, where oxygen is directly conjugated to the ring, has a drastic effect on the C O bond structure, which becomes very much like that of the phenoxy1 C O bond ( v , ~ 1500 The radical anions of terephthalic acid have 11 prr electrons while the p-benzosemiquinone anion has only 9. so it is difficult

-

5437

to compare the nature of the electronic transitions in these types of radicals. However, there is some similarity in the enhancement pattern of the phenyl vibrations in the 530-nm Raman spectra of terephthalic acid radical anions and the 430-nm spectrum of p-benzosemiquinone anion. The vga, vga, and v1 vibrations appear at almost similar relative intensities. 11,12 The 360-nm transition in the terephthalic radical anions is probably similar in this respect to the 3 15-nm transition of the p-benzosemiquinone anion, although due to the lack of the resonance Raman data on the latter radical, a direct comparison is not possible.

Temperature and Solvent-Polarity Dependence of the Absorption and Fluorescence Spectra of a Fixed-Distance Symmetric Chlorophyll Dimer S. G . Johnson, G . J. Small,* Ames Laboratory-US. Iowa 50011

Department of Energy and Department of Chemistry, Iowa State University, Ames,

D. G . Johnson, W. A. Svec, and M. R. Wasielewski Chemistry Division, Argonne National Laboratory, Argonne, Illinois 60439 (Received: January 23, 1989)

Temperature-dependentabsorption and fluorescencespectra and line-narrowed fluorescence and excitation spectra are reported for a dimer consisting of two methylpyrochlorophyllide a molecules that share a vinyl group at the 2-position of each macrocycle. The data (obtained for three solvents of widely differing polarity) show that the dimer exists in two conformations (A and B) and that excited-state relaxation from A to B onsets near the glass transition temperature ( T J . Molecular modeling suggests that the two conformations are related by "bicycling" of the two single bonds joined to the vinyl group linkage. At sufficiently low temperature, the solvent dynamics are rate limiting for the conformational relaxation. For a solvent of sufficiently high polarity (DMF), the excited state of B is shown to access a new radiationless decay channel for T R Tg A charge-transfer state is suggested to be important for this decay. The model presented is shown to provide a qualitative explanation for the frequency domain and recently obtained picosecond and fluorescence quantum yield room-temperature data.

I. Introduction Interest in the primary charge separation process and the excited electronic state structure of photosynthetic reaction centers (RC) has, for many years, stimulated studies of model dimeric and oligomeric chlorophyllic systems. In the case of bacterial RC, the importance of a special bacteriochlorophyll (BChl) pair for the initial phase of charge separation was established prior to the structure determinati~nsl-~ for Rhodopseudomonas viridis and Rhodobacter sphaeroides. The diversity of model systems constructed by covalent linkage, nucleophilic linkage, and self-aggregation is impressive, and their study has provided valuable insight on singlet and triplet excitation and charge delocalization.4-6 In what follows, we mention only a few of the model systems studied. Fong proposed' a dimer of Chl a.H20 as a model for the primary electron donor (PED) of photosystem I (P700) whose assembly is dependent on solvent conditions. Shipman et aL8 presented a similar model for P700 involving dimers formed by nucleophilic linkage of Chl a monomers. Nucleophiles employed (1) Deisenhofer, J.; Epp, 0.;Miki, K.; H u h , R.; Michel, H. J. Mol. Biol. 1984, 180, 385. (2) Allen, J. P.; Feher, G.; Yeates, T. 0.;Rees, D. C.; Deisenhofer, J.; Michel. H.: Huber. R. Proc. Natl. Acad. Sci. U.S.A. 1986. 83. 8589. (3) Chang, C. H.; Tiede, D.; Tang, J.; Smith, U.; Norris; J.; Schiffer, M. FEES Lett. 1986, 205, 82. (4) Katz, J. J.; Hindman, J. C. In Biological Events Probed by Ultrafast Laser Spectroscopy; Alfano, R. R., Ed.; Academic Press: New York, 1982; p 119. (5) Wasielewski, M. R. In Light Reaction Path of Photosynthesis; Fong, F. K., Ed.; Springer-Verlag: West Berlin, 1982; p 234. (6) Boxer, S. G. Biochim. Biophys. Acta 1983, 726, 265. (7) Fong, F. K. Proc. Natl. Acad. Sci. U.S.A. 1974, 71, 3692. (8) Shipman, L. L.; Cotton, T. M.; Norris, J. R.; Katz, J. J. Proc. Natl. Acad. Sci. U.S.A. 1976, 73, 1791.

0022-3654/89/2093-5437$01.50/0

were ROH, RSH, or RNH2, where R is an alkyl chain. A model for the PED state of Rb. sphaeroides, P870, suggested by Wasielewski et aL9 is a pair of bacteriochlorophyllide a molecules covalently bound by an ethylene glycol diester bridge. A similar model for P700 was also constructed by Wasielewski et a1.I0 in which two Chl a were linked by the same bridge. Boxer and Closs" presented a P700 model that is the ethylene glycol diester of methylpyrochlorophyllide a. Pellin et al.l2 have synthesized a P700 electron-acceptor model system consisting of an ethylene glycol diester of pyrochlorophyll a and two molecules of pyropheophorbide e t h y l e n e glycol monoester. The pyropheophorbide a molecules acted as electron acceptors for the dimer. Yuen et al.13 have constructed a model for a P700-antenna system with a tris(pyrochlorophy1lide a)-tris(hydroxymethy1)ethane triester. Two of the molecules folded upon one another when the solvent conditions were sufficient with the third molecule acting as an antenna. Wasielewski5 proposed a model for P700 consisting of a bis(chlorophyl1 a ) cyclophane. Bucks et a1.I4 have investigated several dimers of chlorophyllic species (pyropheophorbide a, pyropheophytin a) and trimers as models for primary electron donors and primary electron donor-antenna systems. Agostiano et a l l 5 (9) Wasielewski, M. R.; Smith, U. H.; Cope, B. T.; Katz, J. J. J . Am. Chem. SOC.1977, 99,4172. (10) Wasielewski, M. R.;Studier, M. H.; Katz, J. J. Proc. Natl. Acad.Sci. U.S.A. 1976, 73, 4282. (11) Boxer, S . G.; Closs, G. L. J . Am. Chem. SOC.1976, 98, 5406. (12) Pellin, M. J.; Wasielewski, M. R. Nature 1979, 278, 54. (13) Yuen, M. J.; Closs, G. L.; Katz, J. J.; Roper, J. A,; Wasielewski, M. R.; Hindman, J. C. Proc. Natl. Acad. Sci. U.S.A. 1980, 77, 5598. (14) Bucks, R. R.; Netzel, T. R.; Fujita, I.; Boxer, S. G. J . Phys. Chem. 1982, 86, 1947. (15) Agostiano, A.; Butcher, K. A.; Showell, M. S.; Gotch, A. J.; Fong, F. K. Chem. Phys. Lett. 1987, 137, 37.

0 1989 American Chemical Society

Johnson et al.

5438 The Journal of Physical Chemistry, Vol. 93, No. 14, 1989

'i0

Figure 1. Methylpyrochlorophyllide a

dimer.

have proposed a model for P680 involving a dimer of Chl a.2H20. Speculation concerning the geometry of the special pair in the RC of Rps. viridis and Rb. sphaeroides ended with the structure determination of the former by Deisenhofer et al.' and of the latter by Chang et aL3 and Allen et aL2 The two BChl molecules have an edge-to-edge conformation with n-a overlap of the two macrocycles restricted to ring I and an angle between the two planes of about 15O. The Q,-transition dipoles of the monomer are out of phase (antiparallel), and from the geometry, it is clear that the lowest energy miniexciton state (denoted as P)is allowed and is antisymmetric with respect to the C2-symmetry axis. From the structures, the importance of protein-pigment interactions for the special pair geometry is apparent. With the above structural features in mind, Johnson et aLi6 synthesized the fixed-distance symmetric pyrochlorophyllide a dimer shown in Figure 1. The molecule has approximate C, symmetry, and the double bond serves to hold the two macrocycles at a constant edge-to-edge distance of approximately 3.9 A. The Mg-Mg distance is about 12 A. The 'H N M R data (nuclear Overhauser effect) indicate the structure shown in Figure 1 has a dihedral angle between the olefinic linkage and the macrocycles of 49-53O. The possibility that this structure is an average for conformers two (or more) rapidly interconverting (at room could not be excluded. The room-T absorption and fluorescence spectra of the dimer reportedi6 are interesting in that the lowest energy Q,-absorption and fluorescence maxima depend only weakly on solvent polarity while the fluorescence lifetime and fluorescence quantum yield exhibit a 2 order of magnitude decrease as the solvent dielectric constant is increased from 2.4 (toluene) to 36.7 (DMF). The absorption and fluorescence maxima occur at -690 and 720 nm at room T . Attempts to observe new emission that accompanies the diminuation of the 720-nm fluorescence intensity were unsuccessful. Johnson et a1.16 also demonstrated, on the basis of stimulated emission studies (610-nm pumping), that the 720-nm emission exhibits a rise time that decreases from 18 ps for toluene to S1.5 ps for DMF. N o prompt stimulated emission signal was observed. In the absence of firm evidence to the contrary, Johnson et aLi6 assumed that the dimer exists in only one ground-state conformation for consideration of dynamical models for interpretation of the time domain and fluorescence quantum yield data. Each model presented certain difficulties as will be briefly discussed in later text. With this in mind, we recently performed T-dependent absorption, fluorescence excitation, and frequency domain fluorescence studies of the pyrochlorophyllide a dimer in order to further explore the nature of the absorbing and emitting species and their electronic states and how they depend on solvent polarity. These studies included fluorescence line-narrowing measurements. The new data establish that the dimer exists in two conformations for both the ground and excited electronic states. The T dependences of the absorption and fluorescence spectra establish that interconversion from the higher energy absorbing conformer to the lower energy conformer onsets in the excited state near the glass transition temperature (T,) for all solvents utilized. The T dependence of the 720-nm fluorescence intensity in different solvents provides an obvious connection with the time domain data (16) Johnson, D. G.; Svec, W. A,; Wasielewski, M. R. I s r . J . Chem., in

press.

TABLE I: Absorption and Fluorescence Maxima Tg, T,,,,' abs max,b fl max: solvent ( E ) K nm nm fl fwhm," cm-' toluene (2.4) 117, 118 685,705 690,722 400,650 (3.5:l) MTHF (7.0) 95, 136 684, 702 690,720 440,550 (2.5:l) DMF (36.7) 129, 212 680,695 685,725 360, 560 (5:l)

Tg and T,,, are the glass and melting transition temperatures. bAbsorption maxima in the low-T limit. At room T, the single absorption, maximum in toluene, MTHF, and DMF occurs at 686,684, and 683 nm, respectively. cFluorescence maxima in the low-T limit. dFull width at half-maximum for the 685- and 720-nm fluorescence bands. The ratio in parentheses is the 720:685integrated fluorescence intensity ratios obtained with A,, = 632.8 nm at low T.

of Johnson et aLi6 This fluorescence is assigned to the lower energy absorbing conformer. A particular intradimer coordinate for conformer interconversion is suggested on the basis of molecular modeling. A two-conformer excited-state model embellished with a charge-transfer (CT) state is shown to provide a consistent, albeit qualitative, interpretation of the time and frequency domain data. The CT state provides for decay of the 720-nm emitting state only in solvents of sufficiently high polarity and for T 2 Tg. The possibility that the C T state is a TICT (twisted intramolecular charge-transfer) state is considered. 11. Experimental Section

The covalently linked dimer of methylpyrochlorophyllide a was prepared and purified by the procedure of Johnson et a1.16 The solvents used for the experiments were purified by standard methods. The 2-methyltetrahydrofuran (MTHF) was treated with lithium aluminum hydride, then distilled, and stored over molecular sieves (3A or 4A). The toluene was distilled once, stored over sodium/benzophenone (a drying agent), then vacuum distilled, and stored over molecular sieves under nitrogen. The dimethylformamide (DMF) was treated with neutral alumina, then distilled, and stored over molecular sieves. Each solvent was freshly prepared before being used. To each solution was added 1% pyridine (by volume; certified ACS grade) to preclude aggregation M based of the dimer. Dimer concentrations used were 1 X on an e = 50 000 for the Qyabsorption band. The samples were sealed in glass tubes of -3-mm i.d. under 4-5 cm of N2(g) to prevent contaminalion due to water. Absorption spectra were obtained with a Bruker IFS 120 H R Fourier transform infrared (visible) spectrometer. The resolution was 4 cm-', and each absorption spectrum presented is the average of 50 scans. Fluorescence spectra were obtained from conventional setups in which either an excimer-pumped dye laser or C W He-Ne laser was used as an excitation source. The former, Lambda Physik FL2002 dye laser pumped by a Lambda Physik EMGlO2 excimer, was also utilized for fluorescence excitation measurements. Fluorescence was dispersed by either a 1-m McPherson 2061 ( R = 160000) or a 1-m Jarrel-Ash ( R = 60000) monochromator. For CW excitation, a PMT and picoammeter were used for signal processing while, for pulsed excitation, either a Tracor Northern TN-6134 intensified diode array and TN-6500 O M A or PMT plus EG & G Model 162 boxcar averager (with two Model 164 processor modules) was employed. The PMT (RCA 31034 or EM1 9558-B) was cooled. Spectra obtained with the pulsed laser were normalized for pulse intensity jitter. A Model 8DT Janis liquid-helium cryostat (convection cooling) was used for T-dependent studies between 4.2 and 300 K. Temperature measurements were made with a Lakeshore Cryotronics DTC-500 K calibrated silicon diode.

-

-

111. Results

The dimer absorption exhibits a marked dependence on the temperature ( T ) . At room T , the absorption appears as a single band near 690 nm while at low T it appears as a doublet with components near 685 and 705 nm. The band positions exhibit a slight dependence on solvent, Table I. Temperature-dependent spectra are shown in Figure 2 for toluene. The spectra obtained for T < 160 K are essentially identical with the 160 K spectrum

Spectra of Fixed-Distance Symmetric Chlorophyll Dimer A h ) 0.12

700

600

8

A

2

0.06

TzI60K

17

3

16000

15000 E (cm-')

14000

13000

Figure 2. Temperature dependence of the dimer absorption in toluene. Spectra are offset for clarity, but integrated intensities may be directly compared. Spectra for T < 160 K do not exhibit further change.

in Figure 2. The spectra in Figure 2 are offset for inspection, but their integrated intensities can be directly compared. In so doing, it is found that the integrated absorption intensity is independent of T. The absorption band for the methylpyrochlorophyllide a monomer is independent of solvent polarity and located at 660 nm.I6 Thus, the dimer bands are significantly red shifted ( 2 2 5 nm) relative to that of the monomer. In what follows, the two dimer components observed in absorption at low Twill be referred to as the 685- and 705-nm bands. The T dependence of the dimer fluorescence excited at 632.8 nm (He-Ne laser) has also been determined for three solvents (toluene, MTHF, DMF). The excitation provides a vibrational excess energy for the 685- and 705-nm absorption components of 1200 and 1600 cm-I, respectively. The excitation energy is too low to provide significant population of the dimer Q,states at -585 nm.I6 Spectra for M T H F and D M F are shown in Figures 3 and 4, respectively. The intensity distributions of the spectra within a given figure can be directly compared. The T dependence of the dimer fluorescence in toluene is similar to that in MTHF. For all three solvents, two fluorescence bands at -685 and -720 nm are readily observed for T 5 T f (freezing point of the solvent). Only the 720-nm fluorescence band is observed for T > T p Thermal cycling experiments showed that the T dependence exhibited in Figures 3 and 4 is reversible. The narrow-line structure seen superimposed on the 685-nm fluorescence band in the lowest T spectra of Figures 3 and 4 is a consequence of fluorescence line narrowing (FLN)." It is firmly established that narrow-line laser excitation into a vibronically congested region of the lowest excited singlet state (SI)can produce vibrationally relaxed and line-narrowed fluorescence bands in the region of the SI So absorption origin (685-nm region for the case at hand). The displacements of the line-narrowed bands from the excitation frequency provide the excited-state vibrational frequencies. Vibronically excited FLN spectroscopy has been applied to a variety of biomolecules, including photosynthetic pigmentsl*-20 and DNA-carcinogen adducts.21 Figure 5 shows the FLN spectrum obtained with A,, = 632.8 nm. Several of the

-

-

-

(1 7) Personov, R. I. In Spectroscopy and Excitation Dynamics of Condensed Molecular Systems; Agranovich, V. M., Hochstrasser, R. M., Eds.; North-Holland: Amsterdam, 1983; p 5 5 5 . (18) Avarmaa, R. A.; Rebane, K. K. Specrrochim. Acta 1985,41A, 1365. (19) Flinfschilling, J.; Williams, D. F. Photochem. Phorobiol. 1977, 26, 109. (20) Avarmaa, R. A.; Rebane, K. K. Chem. Phys. 1982,68, 191. (21) Jankowiak, R. J.; Cooper,R. S.; Zamzow, D.S.; Small, G. J.; Doskocil, G.; Jeffrey, A. M. Chem. Res. Toxicol. 1988, 1 , 60.

The Journal of Physicnl Chemistry, Vol. 93, No. 14, 1989 5439 TABLE 11: Vibrational Modes for the 685-nm Dimer State excited-state freq," cm-l

ground-state freq,b cm-'

298 w 313 vw 345 vs 364 m 376 m 384 m 563 w 570 s 595 m 617 vw 688 w 711 w 728 vw 155 s 773 w 804 m 842 m 868 m 888 m 903 s 918 vs 931 s 953 vw 990 vs 1002 m 1011 w 1025 m 1056 s 1083 m 1098 w 1121 m 1148 s 1169 m 1181 s 1195 w 1244 s 1249 s 1270 s 1288 s 1317 s 1383 m 1402 m 1565 w

307 317

excited-state freq,' cm-I

348 s 358 390

390 w-br

572 603

570 w 600 vw 688 vw

703 733 753

748 s 765 vw

803 85 1

918

985

890 vw 910 vw 925 w 984 vs 1005 vw 1030 vw

1046 1075 w-br 1115 1147 1162 1188

1266 1290 1312 1380 1392 1560

1110 w 1135 w 1168 w 1195 vw 1243 sh 1250 vs 1275 vw 1325 vw 1372 w 1395 vw

Excited-state vibrational frequencies this work. Relative intensities: s = strong, m = medium, v = very, br = broad, sh = shoulder, w = weak. *From resonance Raman of pyrochlorophyll a oligomers. See ref 24. CExcited-statevibrational frequencies of Chl a monomers. See ref 20.

bands are labeled with their associated excited-state vibrational frequencies (cm-I). As expected, the most intense features in Figure 5 are located in the vicinity of 1200 cm-l since this is the excess vibrational energy produced in the 685-nm state with hX = 632.8 nm; vide supra. FLN spectra (not shown) obtained with several other ,A values have been obtained and provide, together with Figure 5, a rather complete picture of the fundamental vibrations for the 685-nm state. These frequencies are in good agreement with those obtained from line-narrowed fluorescence excitation spectra. An example of such a spectrum is shown in Figure 6 for fluorescence detection at 685.0 nm with a 2-cm-' band-pass; cf. section 11. Several of the zero-phonon excitation bands are labeled with a value (cm-I) corresponding to the excited-state vibrational frequency. Table I1 (first column) lists the frequencies and relative intensities of the modes belonging to the 685-nm dimer state provided by the FLN and line-narrowed fluorescence excitation spectra. Also presented in Table I1 are excited-state vibrational frequencies for Chl n in diethyl ether glass at 5 K (third column)I8 and ground-state vibrational frequencies for oligomers of pyrochlorophyll a at 35 K (second column).** Reasonable agreement exists between the vibrational frequencies (22) Lutz, M. In Advances in Infrared and Raman Spectroscopy; Clark, R. J. H., Hester, R. E., Eds.; Wiley Heydon: London, 1984; Vol. 11, p 21 1.

The Journal of Physical Chemistry, Vol. 93, No. 14, 1989

5440

Johnson et al.

;A!A

A

g

g-

-

C

c3

C r+

D-i

LL

U

LL

-

-

2

L

c

E

-

-

640

600

720

760

000 820

640

600

720

760

n ]

800 830

640

000 830

T1270K

YL3

C

-

-

x. 4

U 640

U

720 760 X(nm)

760

VI

r+

600

720

-

.640

680

.-C

T = l38K

n

T=232K

8 0 0 030 X(nm)

680

720

760

000 030

X(nm)

Figure 3. Temperature dependence of the dimer fluorescence in MTHF. A,, = 632.8 nm, IO-cm-' resolution. Integrated intensities for different temperatures may be directly compared.

Lo

r.

---

z

C VI C

0)

aJ C

than 632.8 nm. The fluorescence observed appeared as it does with A, = 632.8 nm; cf. Figures 3 and 4. In order to determine whether the 720-nm fluorescence is connected with the 705-nm absorption band, fluorescence excitation spectra were obtained as a function of temperature for a detection wavelength of 730 nm (band-pass = 30 cm-I). Figure 7 shows the dependence of the fluorescence excitation band maximum on temperature for toluene. Also shown in this figure is the dependence of the 720-nm fluorescence band position on temperature. For both sets of data, a significant band shift occurs near 180 K, which is close to the freezing point of toluene (Tf = 178 KZ3). From Figure 7, it can be seen that the low- and high-T values for the fluorescence excitation band maximum are -706 and 694 nm. This 12-nm shift is the same as that observed for the 720-nm fluorescence band within experimental uncertainty. This together with the similarity between the two solid curves in Figure 7 indicates that the 720-nm fluorescence originates from the 705-nm-absorbing dimer state. Thus, this state is characterized by a Stokes shift of 15 nm (- 300 cm-'), far larger than the Stokes shift for the 685-nm dimer state. These two dimer states are further contrasted by the facts that the maximum of the 685-nm band is only weakly dependent on the temperature and does not undergo an abrupt change at T Tf.

-

VI

g U 640 660

600 700

720

740 760 X(nmi

780 000

030

Figure 4. Temperature dependence of the dimer fluorescence in DMF: A, 5 K; B , 9 1 K ; C , 140 K; D, 176 K ; E , 201 K ; a n d F , 2 7 0 K. See

Figure 3 caption. and relative intensities for the dimer with those from these references. The FLN and line-narrowed excitation spectra establish that the 685-nm fluorescence originates from the 685-nm-absorbing dimer state. The small Stokes shift observed for the 685-nm state is consistent with small Franck-Condon factors for both the intramolecular and phonon modes. In addition, these spectra establish that the width of the 685-nm absorption band at low Tis dominated by site inhomogeneous line broadening. This has also been confirmed by the nonphotochemical hole-burned spectra for the 685-nm band (spectia not shown). In contrast with the 685-nm emission, FLN for the 720-nm emission was not observed even with &, values significantly higher

-

-

IV. Discussion A. Temperature-Dependent Absorption and Fluorescence Spectra. The results of the preceding section establish that the 685- and 705-nm absorption bands are due to the two electronic states responsible for the 685- and 720-nm fluorescence bands. (23) Weast, R. C., Ed. CRC Handbook of Chemistry and Physics; CRC Press: Boca Raton, FL, 1979.

The Journal of Physical Chemistry, Vol. 93, No. 14. 1989 5441

Spectra of Fixed-Distance Symmetric Chlorophyll Dimer 7

I

I

I

724

'

I

'

. I

'

I

'

r

W W

'u A

.-cC

u)

3

F

.-e e a e

Y

> k v)

z W

I-

z,

w

u

z

m m

W

W

u

rn

0

W

a

5

160 240 TEMP. ( K )

3:

Figure 7. Temperature dependence of the fluorescence (720-nm) and photoexcitation (observation at 730 nm) band position in toluene.

-I LL

-

660

700

680 X(nm)

Figure 5. Fluorescence line-narrowed spectra associated with the 685-nm emission. In toluene, 5 K, 8*cm-' resolution.

0

f r ??

630

635

640

645

650

655

660

X(nm) Figure 6. Line-narrowed fluorescence excitation spectra. Observation at 685 nm, T = 5 K, 2-cm-' resolution.

The possibility that one of the two absorption bands is due to an impurity (monomer or dimer) can be ruled out since the HPLC and ' H N M R analyses of the extensively purified dimer yield a limit for an impurity concentration that is no greater than 5%.16 D e c o n v ~ l u t i o nof~ ~the low-T absorption spectrum in Figure 2 yields a 685:705 integrated intensity ratio of about unity. This is inconsistent with either of the two components being due to an impurity (at a level of Tf is compensated by a gain in the 720-nm fluorescence intensity; vide infra. This is not consistent with the behavior of an isolated impurity. The T-dependent fluorescence behavior also argues against the 705-nm absorption band being due to an aggregate(s) of the dimer (analogous to aggregates of Chl monom e r ~ ~ ~Furthermore, ,~~). the solvent purification procedure and the addition of 1% pyridine (an excellent ligand for Mg) should preclude aggregate formation especially since dilute solutions (- 10-5 M) were used in the experiments. Typically, aggregation requires much higher concentrations, on the order of 10-2-10-3 M.25-27 We proceed now to consider the possibility that a single dimer conformation (e.g., as in Figure 1) can account for the Tdependent absorption and fluorescence data. To do so requires a consideration of the exciton model for the dimer whose constituent monomers possess Qy transition moments that are antiparallel (180' out of phase). We denote the two dimer states by Pi (plus and minus indicating symmetric and antisymmetric with respect to the dimer symmetry axis) and the Davydov splitting by 2K In the absence of symmetry breaking (e.g., due to the solvent), only the P state carries oscillator strength, and furthermore, it is this state that lies lower in energy since V > 0.28 Clearly, the presence of two absorption bands cannot be reconciled with this model. Consider next that solvent-induced symmetry breaking scrambles P+ and P- to an extent that endows them with comparable oscillator strengths or that the conformation of the dimer is not that shown in Figure 1. This single conformation would be one for which the monomer transition dipoles are not antiparallel (or parallel) so that both P+ and P- carry oscillator strength.29 There are several difficulties with both of these possibilities. First, it is well documented for organic crystals relaxation from an upper exciton (Davydov) level to a lower energy exciton level m u r s on

-

(25) Amster, R. L. Photochem. Photobiol. 1969, 9, 331. (26) Cotton, T. M.; Loach, P. A,; Katz, J. J.; Ballschmiter,K. Photochem. Photobiol. 1978, 27, 735. (27) Brody, S . S.; Broyde, S. B. Biophys. J . 1968,8, 1511. (28) Calculations performed with the dipoledipole approximation yielded V = 120 cm-I for the conformation in Figure 1. (29) Davydov, A. S. Theory of Molecular Excitons; Plenum: New York, 1971. . (3O),Craig, D. P.; Walmsley, S. H. Excitons in Molecular Crystals; Benjamin: New York, 1968. (31) Port, H.; Rund, D.; Small, G. J.; Yakhot, V. Chem. Phys. 1979, 39, 175 - . _.

(32) Johnson, C. K.; Small, G. J. In Excited States; Lim, E. C., Ed.: Academic Press: New York, 1982; Vol. 6, p 97.

5442

The Journal of Physical Chemistry, Vol. 93, No. 14, 1989

/

\ \

/

\ \

/

/

‘--’

Johnson et al. B emits at 720 nm. Furthermore, the 720-nm fluorescence band is present at all temperatures. All three factors indicate that both A and B are stable conformers at room T . The low-temperature absorption spectra then establish that conformers A and B have comparable concentrations (making the reasonable assumption that their oscillator strengths are similar). Deconvolution of the low-T absorption profiles in terms of two Gaussians showed that 685- and 705-nm absorption bands are equal in intensity (&30%) for toluene, MTHF, and DMF. However, the fluorescence spectra obtained with X, = 632.8 nm show that the 720-nm fluorescence is 3-5 times more intense than the 685-nm fluorescence at low T , Table I. This excitation wavelength should excite A and B with nearly equal probability, Figure 2. Obvious (but unsubstantiated) arguments involving differences in oscillator strengths and radiationless decay constants for A and B could be presented to explain the greater fluorescence intensity from B. We consider now the T dependence of the 685-nm fluorescence band. For all solvents, the intensity was constant from T 5 K to Tg,where Tg is the glass transition temperature, Table I. As the temperature was increased above Tg, the fluorescence intensity rapidly diminished. By T Tf (Table I), the 685-nm fluorescence was not detectable. As discussed earlier, the diminution of this fluorescence for toluene and M T H F is compensated by a gain in the intensity of the 720-nm emission. Thus, it appears that large-amplitude solvent motion is required for relaxation from the 685-nm state of A to the 705-nm state of B. If one assumes that for T > T, the two conformers in their excited state are in thermal equilibrium, the Tdependence of the 685-nm fluorescence intensity leads to A’ = 360 and 475 cm-’ for MTHF and DMF, respectively; cf. Figure 8. Considering for the moment that in Figure 8 A = 0 and noting that the 685- and 705-nm absorption bands are separated by -400 cm-l, it follows that A’ = 400 + ’/* Stokes shift for the case where w’ w , Figure 8. The Stokes shift is -300 cm-’, Table I, so that a calculated value for A’ = 550 cm-’. Improved agreement with the experimental values for M T H F and D M F can be achieved by allowing for A < 0 and/or w’ C w. Indirect evidence for w’ < w is that the low-T fwhm of the 705-nm absorption band (from deconvolution) is about 100 cm-I narrower than the fwhm of the 720-nm fluorescence band in all solvents used. To conclude this section, we consider the T-dependent fluorescence spectra for D M F shown in Figure 4. Although the behavior for the 685-nm fluorescence band is similar to those for toluene and MTHF, the T dependence of the 720-nm band is distinctly different. That is, the latter band undergoes a significant decrease in intensity as T is increased above i= 120 K ( Tg = 129 K). By room T, its intensity decreased by about a factor of 10. This interesting behavior is best discussed in the context of the recent time-domain studies, section IVC. B. Possible Dimer Conformations. The data presented in text above indicate that relaxation from conformer A to B occurs with a rate competitive with fluorescence as T approaches Tg. At T Tf,the relaxation rate has significantly increased to the point where fluorescence from A is no longer observable. Detailed considerations of the dimer structure, Figure 1, indicate that the most plausible intradimer vibrational coordinate for conformational relaxation is “bicycling” by the two single bonds joined to the vinyl group linkage. The dihedral angle between the olefinic linkage and the macrocycles shown in Figure 1 is -5OO. A computer molecular model program (BIOGRAF) was used to gauge the steric limitations imposed by the a,vinyl, and 1-methyl protons on bicycling. The 12, 2, and ethylenic carbons were used to define the dihedral angle discussed throughout this paper (the chlorophyll carbons being numbered in the standard manner4). It was found that dihedral angles between -50’ and 140° satisfied the van der Waals requirements for the above protons with the macrocyclic planes in a parallel configuration. It is important to note that *-electron delocalization between the two macrocycles via the vinyl group would be minimized for the 90’ position and that this configuration could lead to formation of a TICT state (twisted intramolecular charge-transfer state). We return to this point in section IVC but note now that such a state could be expected

-

-

QFigure 8. Hypothetical potential energy curves for two conformers of the

dimer. a picosecond or subpicosecond time scale even at very low T. Very recently, a decay time of 260 fs at 1.6 K has been measured for an upper exciton level of the 7-bacteriochlorophyll a containing protein subunit of the light-harvesting complex from Prosthecochloris ae~tuarii.’~There is no precedent for the suggestion that the observation of the 685-nm fluorescence at low T i s due to very slow (approximately nanosecond) decay of the 685-nm exciton level to the 705-nm level. Second, theobservation that the 685-nm Tfis difficult to reconcile level abruptly begins to decay a t T in terms of existing theories for interexciton level relaxation. Third, if the 685- and 705-nm absorption bands are exciton components of a single dimer configuration, then it is necessary to explain why the 705- and 685-nm levels shift and do not shift (respectively) appreciably with temperature. That is, the Davydov splitting is T dependent but the energy of the upper component is T independent. This could occur if the expected shift for the upper component was canceled by the T dependence of the dispersion energy. However, this possibility is viewed as extremely remote. Thus, we conclude that the T-dependent fluorescence and absorption data cannot be understood in terms of a single conformation of the dimer shown in Figure 1. Consequently, a model in which two stable conformations (A and B) of the dimer exist is explored next. The 685- and 705-nm absorption bands will be considered to be the Q,-origin bands of A and B. Possible conformer structures will be considered later, section IVB. Hypothetical potential energy curves for the ground and excited states of the dimer are given in Figure 8 (solid curves). The potential well minima for conformer B are shifted relative to each other because of the Stokes shift associated with the 705-nm state. For each of the solvents used, the integrated absorption intensity remains constant as the temperature increases from the low-T limit (where the 685- and 705-nm components are resolved) to room T (where only a single absorption band is apparent). The spectra for all solvents, e.g., Figure 2, show that as T increases past Tf, the 705-nm band blue shifts until it merges with the 685-nm band. The data in Figure 7 for toluene indicate that a t room T the 705-nm band has shifted to -694 nm. Within our experimental uncertainty, the 685-nm band does not appear to shift with temperature. It was argued above that the 705-nm state or conformer

-

-

(33)Johnson, S . G.;Small, G. J. Chem. Phys. Lett. 1989, 155, 371. (34)Craig, D. P.;Dissado, L. A. Chem. Phys. Lett. 1976, 44, 419. (35)Clark, M.; Craig, D. P.; Dissado, L. A. Mol. Cryst. Liq. Cryst. 1978, 44, 309. (36)Toyozawa, Y. J . Lumin. 1970, 112, 732. (37) Yarkony, D.; Silky, R. J . Chem. Phys. 1976.65, 1042. (38) Freed, K.J . Am. Chem. SOC.1980, 102, 3130.

-

-

-

Spectra of Fixed-Distance Symmetric Chlorophyll Dimer to experience a significant stabilization in a high-polarity solvent such as DMF. Calculations of the potential energy minima along the bicycling coordinate would require consideration of atom-atom interactions and the 7-electron stabilization energy associated with partial conjugation across the bridge between the two macrocycles (for. both the ground and excited electronic states). In the absence of such calculations, it is, nevertheless, clear that more than one thermally accessible conformation is entirely plausible, consistent with the conclusion of the preceding section that the dimer exists in two conformations (A and B) in all solvents utilized. The correlation between the onset of relaxation of conformer A with the glass transition temperature, Tg,is convincing evidence for the importance of large-amplitude solvent motion for configurational change in the dimer. However, the excited-state energies of A and B (685- and 705-nm states) and the Stokes shift for B are not very sensitive to solvent polarity, Table I. But, again, solvent polarity has a profound effect on the radiationless decay of the 705-nm state at higher temperatures. C . Radiationless Decay of the 705-nm State of Conformer E . The data presented here establish that, for a solvent of sufficiently high polarity such as D M F ( t = 37), an efficient radiationless decay channel for the 705-nm state of conformer B (emission at 720 nm) is accessed. This observation provides a connection with the picosecond transient absorption, fluorescence quantum yield, and stimulated emission studies of Johnson et a l l 6 (referred to hereafter as JSW), which were performed at room T with a series of solvents of different polarity. At room T , JSW were only able to monitor the 720-nm emission band. They observed a 2 order of magnitude decrease in fluorescence quantum yield from toluene ( t = 2, q$, = 0.18) to D M F (e = 37, &, = 0.002), A,, = 425.0 nm. A stimulated emission feature was observed in the transient absorption spectra that coincides with the 720-nm fluorescence maximum. Excitation was at 610 nm (an important fact for what follows). The stimulated emission observed by JSW exhibited a rise time ( T ~ )that was solvent dependent, varying from 51.5 ps for D M F to 18 ps for toluene (the least polar solvent used). In what follows, the two-conformer model for the dimer suggested in this paper is shown to provide a qualitative explanation for the data of JSW when relaxation of the 705-nm state into a charge-transfer (CT) state is invoked. Such a state was discussed by JSW. However, in view of the data presented here, this model was realized to present certain difficulties. For example, the rise time ( T J was assumed to arise from the decay of the upper exciton level of the dimer shown in Figure 1 to the dipole-allowed lower exciton level (P-). The former (P') is dipole forbidden, in the absence of solvent perturbations, and it was difficult to understand why excitation at 610 nm should preferentially excite P+ rather than P.Excitation of the latter would lead to prompt stimulated emission (excitation at 610 nm is lower than the onset of absorption due to Q,transitions). No prompt stimulated emission was detected.I6 Within the two-conformer (685, 705 nm) model, the T~ data can be understood as follows. The A,, = 610 nm utilized by JSW 1600-cm-I vibrational excitation for conformer A provides absorbing at -685 nm, Table I. From the point of view of maximal vibronic Franck-Condon factors, this is an optimal region for chlorophyllic specie^.'^^^^ On the other hand, an additional -250 cm-' of vibrational energy is provided for the excited state of B, which should lead to a decrease in the Franck-Condon factor by about a factor of Thus, kx= 610 nm preferentially excites conformer A by an amount governed by this factor. Therefore, it may be concluded that the stimulated emission rise time is a direct measure of the rate of relaxation from conformer A to B at room T . To account for the marked dependence of the 720-nm fluorescence lifetime on solvent polarity, we follow JSW and suggest that solvents with sufficiently high polarity stabilize a C T

-

(39) Gillie, J. K.; Small, G. J.; Golbeck, G . H. J . Phys. Chem. 1989, 93, 1620.

The Journal of Physical Chemistry, Vol. 93, No. 14, 1989 5443 state to an extent that places it sufficiently close to the 705-nm state to allow for its decay. Such a C T state is schematically shown in Figure 8 as the dashed potential energy curve. We hasten to add that the C T state need not lie lower in energy than the 705-nm state of B for efficient relaxation of the latter into the former, as recently discussed by Won and Friesner40 in connection with charge separation in photosynthetic reaction centers. On the other hand, the C T state cannot lie too far above 705 nm, as is apparently the case for nonpolar solvents such as toluene. A quantitative discussion of this point requires knowledge of the exchange interaction between the 705-nm state (which may be assumed to be excitonic) and the C T state as well as the relevant electron-vibration coupling strengths of the two states.38 These are not known. However, the weak dependence of the Stokes shift of the 705-nm state on solvent polarity establishes that this shift is primarily a consequence of configurational displacement within the dimer, as suggested by Figure 8. The good correlation between the fluorescence quantum yield and lifetime data presented by JSW establishes that the 685-nm state does not decay directly into the C T state but indirectly via the 705-nm state. On the basis of the electrochemical data presented by JSW, one can roughly estimate that the hypothetical C T state would lie about 1000 cm-l above the 705-nm state. JSW were not able to observe fluorescence from the proposed C T state. This suggests that this state is dark or that it emits in the IR (>900 nm). A possible candidate for the C T state is a TICT state of the type mentioned in section IVB. and associated with the 90' dihedral angle. Very often, TICT states are observed to exhibit fluorescence whose maximum depends sensitively on solvent polarity, viscosity, and t e m p e r a t ~ r e . 4 ~However, ~ ~ ~ there are several examples of compounds for which convincing evidence for a nonemissive TICT state has been In summary, we have established that relaxation from the 685-nm dimer state of conformer A to the corresponding state of B at 705 nm onsets at T = T8 in solvents of widely differing polarity. Thus, large-amplitude solvent motion is required for relaxation from A to B occur on the nanosecond to picosecond time scale. At the same time, the energies of the above two states do not depend strongly on solvent polarity. Molecular modeling suggests that the two conformations of the dimer are related by motion along the bicycling coordinate discussed in section IVB. It is further demonstrated that a radiationless decay channel for B's 705-nm state opens at T = Tg,but only in solvents of sufficiently high polarity, e.g., DMF. The fact that this channel in DMF is closed at T C Tgindicates that solvent motion is required for the formation of a new excited-state dimer configuration, which acts as a precursor for decay into a C T state. If the C T state is a TICT state, this new configuration would be one for which the dihedral angle is ==90° (see section IVB). For this model, the dihedral angle for B would be intermediate in value between 90' and 50' since direct decay from the 685-nm state of A (dihedral angle -50') into the C T state (dihedral angle -90') does not occur. Decay occurs from conformer A to the C T state only through the 705-nm state of B.

V. Concluding Remarks The T-dependent data presented here establish that the methylpyrochlorophyilide a dimer exists in two conformations, A and B. It is suggested that the picosecond stimulated emission (720-nm) rise time ( T ~ datal6 ) reflect excited-state relaxation from conformer A to B at room temperature. To better understand the dependence of this relaxation on the solvent ( T ~= 18 ps for toluene, 51.5 ps for DMF) will require the measurement of T~ as a function of temperature for different pumping wavelengths. (40) Won, Y.; Friesner, R. A. J . Phys. Chem. 1988, 92, 2214. (41) Rettig, W. Angew. Chem., Int. Ed. Engl. 1986, 25, 971. (42) Heldt, J.; Gormin, D.; Kasha, M. Chem. Phys. Lett. 1988, 150, 433. (43) Jones, G.; Jackson, W. R.; Halpern, A. M. Chem. Phys. Lett. 1980, 72. 391. (44) Jones, G.; Jackson, W. R.; Choi, C.-Y.; Bergmark, W. R. J . Phys. Chem. 1985.89. 294. (45) Lipkrt,'E.; Rettig, W.; BonaciE-Koutecky, V.; Heisel, F.; Mieht, J. A. Adu. Chem. Phys. 1987, 68, 1 .

J . Phys. Chem. 1989, 93, 5444-5453

5444

The solvent dependence observed at room T might reflect solvent dynamics or the fact that the barrier height in the excited state, Figure 8, is determined, in part, by solvent-dimer interactions. The observation that the relaxation onsets (becomes competitive with fluorescence) for T = T, proves that solvent dynamics can be expected to be rate limiting at lower temperatures in the range T, 5 T 5 300 K. This need not be the case at room temperature. In addition to the excited Qy states of A and B, a third excited state (CT) has been implicated as a decay channel for the Qy (705-nm) state of B in solvents of sufficiently high polarity. The observation that this channel opens (is competitive with fluorescence) for T = Tg suggests that large-amplitude solvent motion is the first step of the relaxation dynamics. It will be important in future studies to measure the T dependence of the 705-nm state emission (at 720-nm) lifetime in high-polarity solvents. Although the association of the third excited state with a C T state (a TICT state was considered) is reasonable, it is not proven. It is conceivable, for example, that radiationless decay (SI So) from the excited-state conformation of B achieved by solvent motion ( T Z T J would be polarity dependent. The reason why the Stokes shift (linear electron-phonon coupling) is moderately large for conformer B (but small for A) is not clear. The absence of fluorescence line narrowing for B's 720-nm emission is consistent with strong electron-phonon coupling. The Stokes shift, given approximately by 2Sw where w is a mean phonon frequency, is about 300 cm-' in all solvents measured. If we take w 30-50 cm-', then S 5-3. Since the Franck-Condon factor for the zero-phonon transition is exp(-S), S values in this range would render this transition quite highly Franck-Condon forbidden. In future studies, it will be important to perform nonphotochemical hole burning46 on the

705-nm absorption band (and the 685-nm band for comparison) in different glassy solvents in order to determine w and whether this effective mode frequency is associated with an intradimer (e.g., bicycling) or a solvent coordinate. Finally, we comment that while there are some similarities between the spectroscopic, electronic structural, and photophysical properties of the synthetic dimer and the special pair of photosynthetic bacterial reaction centers, the ability of the former to assume different conformations may set it apart from the special pair in which conformational changes are, as yet, unobserved. Nevertheless, the synthetic dimer presents very interesting possibilities for future studies on the static and dynamic roles of solvents in intramolecular conformational changes and the interaction of excitonic and C T states. Acknowledgment. The Ames Laboratory is operated for the U S . Department of Energy by Iowa State University under Contract No. W-7405-Eng-82. The research at the Ames Laboratory was supported by the Director for Energy Research, Office of Basic Energy Sciences. The Argonne work was supported by the Division of Chemical Sciences, Office of Basic Energy Sciences, US. Department of Energy under Contract No. W-31109-ENG-38. S.G.J. acknowledges Daniel S. Zamzow and Dr. Ryszard Jankowiak of Ames Laboratory for technical assistance and Prof. Keith Woo of Iowa State University for helpful discussions regarding structural aspects. Registry No. Methyl pyrochlorophyllide a dimer, 121013-63-6.

--

-

-

(46) Small, G. J. In Spectroscopy and Excitation Dynamics of Condensed Molecular Systems; Hochstrasser, R. M., Ed.; North-Holland: Amsterdam, 1983; p 515.

Interpretation of Temperature-Dependent Photoionization Mass Spectra of +Hexane, Cyclohexane, and Diethyl Ether R. Hoogerbrugge,* M. Bobeldijk, and J. Los FOM-Institute for Atomic and Molecular Physics, Kruislaan 407, 1098 SJ Amsterdam, The Netherlands (Received: March 23, 1988; In Final Form: December 15, 1988)

The mass spectra of vapor-phase n-hexane, cyclohexane, and diethyl ether are measured as a function of temperature by photoionization mass spectrometry. Three fixed wavelengths are used, the Kr I, Ar I, and Ne I resonance lines. The results are interpreted on the basis of a simplified version of the quasi-equilibriumtheory. In this model it is assumed that the density of states of a transition state can be described by the density of states of the neutral molecule multiplied by a phase space scaling factor. The phase space scaling factors are fitted for an optimum reconstruction of the photon and temperature-dependent mass spectra. The knowledge obtained about the fragmentation reaction rates of n-hexane is applied to field ionization mass spectra, which results in an estimate of the average energy deposition in the molecular ion of 0.77 i 0.1 eV.

Introduction The study of temperature-dependent photoionization mass spectra (PIMS) is of both practical and fundamental importance. The practical value is clear since the results show that a good control of the ion source temperature is essential to obtain reproducible mass spectra. In this paper we will concentrate on the fundamental aspects of temperature-dependent PIMS. In the study of the dissociation reaction rates of diatomic molecules, precise knowledge of all the quantum states is necessary. For polyatomic molecules the number of electronic, vibrational and rotational quantum states becomes excessively high; for example, *Towhom correspondence should be addressed at Laboratory of Organic-Analytical Chemistry, National Institute of Public Health and Environmental Protection, Antonie van Leeuwenhoeklaan 9, P.O.Box 1, 3720 Bilthoven, The Netherlands.

0022-3654/89/2093-5444$01.50/0

n-hexane has 57 nuclear and 150 electronic internal degrees of freedom. Due to the large number of degrees of freedom and their strong interaction, the precise information on all the quantum states can never be obtained. On the other hand, the large number of states and their rapid exchange of energy offer the possibility to describe the molecule or molecular ion as a microcanonical ensemble. The statistical description of a single molecule was introduced by Rice and Ramsperger' and by KasselZand extended by M a r c ~ s . The ~ basic assumptions of this RRKM theory, or quasi-equilibrium theory (QET) of mass spectra: are the decay (1) Rice, 0. K.; Ramsperger, H. C. J. Am. Chem. SOC.1927, 49, 1617. (2) Kassel, L. S.J . Phys. Chem. 1928, 32, 1065. (3) Marcus, R. A. J. Chem. Phys. 1952, 20, 359. (4) Rosenstock, H. M.; Wallenstein, M. B.; Wahrhaftig, A. L.; Eyring, H. Proc. Natl. Acad. Sci. U.S.A. 1952, 38, 667.

0 1989 American Chemical Society